Feature Request: Native Support for Z.AI (Zhipu AI) API & Latest GLM Models

[SalimBinYousuf1]

Please add native support for Z.AI (Zhipu AI) API keys in Nanobot, enabling direct access to the latest GLM-4.x and GLM-4.7 series models.

Z.AI is the official platform for Zhipu AI and provides high-performance large language models such as GLM-4.7, GLM-4.6, GLM-4.5, including Flash, Air, and Vision (V) variants. These models offer strong reasoning, coding, agentic workflows, multimodal capabilities, and very long context windows (up to ~200K tokens).

Adding Z.AI support would significantly expand Nanobot’s provider ecosystem and give users access to fast, cost-effective, and widely adopted models.

Why this is valuable

Access to latest flagship models like GLM-4.7 (reasoning, coding, agents)

Support for vision-language models (GLM-4.5V / 4.6V)

Long-context capabilities suitable for large documents and agents

Competitive pricing and strong availability in regions with limited access to other providers

Aligns with Nanobot’s goal of being provider-agnostic

Suggested implementation

Configure via ZAI_API_KEY (env or config file)

Add Z.AI as a selectable provider

Support chat/completions initially (streaming optional)

Model selection examples:

zai/glm-4.7

zai/glm-4.6

zai/glm-4.5-air

zai/glm-4.6v

[pjperez]

I am using it with GLM Coding Plan.

In config.json:

  "agents": {
    "defaults": {
      "workspace": "~/.nanobot/workspace",
      "model": "zai/glm-4.7",
      "max_tokens": 32768,
      "temperature": 0.3,
      "max_tool_iterations": 20
    }

And...

 "providers": {
   "zhipu": {
      "api_key": "xxxxxxxxxxxxxxxxxxxxx",
      "api_base": null
    },

Please make sure you either write zai/glm-4.7 or just glm-4.7, but not zhipu/glm-4.7 as the downstream LiteLLM expects zai/. Nanobot was adding zhipu/ to glm-4.7 so I've sent a small PR to fix that: #32

[anomit]

The above suggestion by @pjperez did not work for me.

nanobot agent -m "What is 2+2?"
Error: No API key configured.

This is the current structure of my config.json

{
  "agents": {
    "defaults": {
      "workspace": "~/.nanobot/workspace",
      "model": "zai/glm-4.7",
      "maxTokens": 32768,
      "temperature": 0.3,
      "maxToolIterations": 20
    }
  },
  "channels": {
    "whatsapp": {
      "enabled": false,
      "bridgeUrl": "ws://localhost:3001",
      "allowFrom": []
    },
    "telegram": {
      "enabled": false,
      "token": "",
      "allowFrom": []
    }
  },
  "providers": {
    "anthropic": {
      "apiKey": "",
      "apiBase": null
    },
    "openai": {
      "apiKey": "",
      "apiBase": null
    },
    "openrouter": {
      "apiKey": "",
      "apiBase": null
    },
    "zhipu": {
      "api_key": "zai-glm-key",
      "api_base": null
    }
  },
  "gateway": {
    "host": "0.0.0.0",
    "port": 18789
  },
  "tools": {
    "web": {
      "search": {
        "apiKey": "",
        "maxResults": 5
      }
    }
  }
}

[pjperez]

Sorry @anomit , may need this #32

[anomit]

Tried running your fork as well @pjperez with the same config

git clone https://github.com/pjperez/nanobot.git nanobot-pjperez
cd nanobot-pjperez
uv venv
source .venv/bin/activate
uv pip install nanobot-ai

Nanobot path shows up as within the venv, as expected

where nanobot

/Users/anomit/workspace/nanobot-pjperez/nanobot/.venv/bin/nanobot

nanobot status
🐈 nanobot Status

Config: /Users/anomit/.nanobot/config.json ✓
Workspace: /Users/anomit/.nanobot/workspace ✓
Model: zai/glm-4.7
OpenRouter API: not set
Anthropic API: not set
OpenAI API: not set

nanobot --version
🐈 nanobot v0.1.0


nanobot
                                                                                                                                                                                        
 Usage: nanobot [OPTIONS] COMMAND [ARGS]...                                                                                                                                             
                                                                                                                                                                                        
 🐈 nanobot - Personal AI Assistant                                                                                                                                                     
                                                                                                                                                                                        
╭─ Options ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ --version             -v                                                                                                                                                             │
│ --install-completion            Install completion for the current shell.                                                                                                            │
│ --show-completion               Show completion for the current shell, to copy it or customize the installation.                                                                     │
│ --help                          Show this message and exit.                                                                                                                          │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭─ Commands ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ onboard   Initialize nanobot configuration and workspace.                                                                                                                            │
│ gateway   Start the nanobot gateway.                                                                                                                                                 │
│ agent     Interact with the agent directly.                                                                                                                                          │
│ status    Show nanobot status.                                                                                                                                                       │
│ channels  Manage channels                                                                                                                                                            │
│ cron      Manage scheduled tasks                                                                                                                                                     │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

No luck yet

nanobot agent -m "what is 2+2"
Error: No API key configured.

[SalimBinYousuf1]

I think base url is not correctly configured ..

[anomit]

Did the oddest workaround ever, playing with a lot of permutations and combinations to get it working.

I added the zai endpoint to mimic anthropic provider under the anthropic provider, as well as appended a zai section by myself.

Added API key in both. If either of the above are left blank or absent, it wont work with the same error "No API key configured".

And even then, nanobot asks for an explicit env to be set with ZAI API key. Which I did, and it works.

nanobot agent -m "What is 2+2?"
2026-02-03 19:00:42.127 | INFO     | nanobot.agent.loop:_process_message:138 - Processing message from cli:user

🐈 Error calling LLM: litellm.AuthenticationError: AuthenticationError: ZaiException - The api_key client option must be set either by passing api_key to the client or by setting the 
ZAI_API_KEY environment variable

export ZAI_API_KEY=<my key>

nanobot agent -m "What is 2+2?"
2026-02-03 19:01:09.515 | INFO     | nanobot.agent.loop:_process_message:138 - Processing message from cli:user

🐈 2 + 2 = 4

nanobot agent -m "what is 10 mod 5"
2026-02-03 19:01:57.849 | INFO     | nanobot.agent.loop:_process_message:138 - Processing message from cli:user

🐈 10 mod 5 = 0

Here is how my config.json now stands. This has been super confusing.

{
  "agents": {
    "defaults": {
      "workspace": "~/.nanobot/workspace",
      "model": "zai/glm-4.7",
      "maxTokens": 32768,
      "temperature": 0.3,
      "maxToolIterations": 20
    }
  },
  "channels": {
    "whatsapp": {
      "enabled": false,
      "bridgeUrl": "ws://localhost:3001",
      "allowFrom": []
    },
    "telegram": {
      "enabled": false,
      "token": "",
      "allowFrom": []
    }
  },
  "providers": {
    "anthropic": {
      "apiKey": "my key",
      "apiBase": "https://api.z.ai/api/anthropic"
    },
    "openai": {
      "apiKey": "",
      "apiBase": null
    },
    "openrouter": {
      "apiKey": "",
      "apiBase": null
    },
    "zai": {
	"apiKey": "my key",
	"apiBase": "https://api.z.ai/api/paas/v4"
    }
 },
  "gateway": {
    "host": "0.0.0.0",
    "port": 18789
  },
  "tools": {
    "web": {
      "search": {
        "apiKey": "",
        "maxResults": 5
      }
    }
  }
}

[pjperez]

@anomit ah! I think I know what the issue was.

LiteLLM expects ZAI_API_KEY and not ZHIPUAI_API_KEY which is what litellm_provider.py tries to se.

nanobot/nanobot/providers/litellm_provider.py

Lines 52 to 53 in d9d744d

	elif "zhipu" in default_model or "glm" in default_model or "zai" in default_model:
	os.environ.setdefault("ZHIPUAI_API_KEY", api_key)

[Myskip]

@anomit ah! I think I know what the issue was.

LiteLLM expects ZAI_API_KEY and not ZHIPUAI_API_KEY which is what litellm_provider.py tries to se.

nanobot/nanobot/providers/litellm_provider.py

Lines 52 to 53 in d9d744d

elif "zhipu" in default_model or "glm" in default_model or "zai" in default_model:
os.environ.setdefault("ZHIPUAI_API_KEY", api_key)

Yes，i solve this problem with ZAI_API_KEY.Here is my config.

{
  "agents": {
    "defaults": {
      "workspace": "~/.nanobot/workspace",
      "model": "zai/glm-4.7",
      "maxTokens": 8192,
      "temperature": 0.7,
      "maxToolIterations": 20
    }
  },
  "providers": {
    "zhipu": {
      "apiKey": "my_api_key",
      "apiBase": "https://open.bigmodel.cn/api/coding/paas/v4"
    }
  }
}

[anomit]

Thanks but why would you post your API key publicly sir?!

On Tue, 3 Feb, 2026, 20:48 Myskip, ***@***.***> wrote: *Myskip* left a comment (HKUDS/nanobot#2) <#2 (comment)> @anomit <https://github.com/anomit> ah! I think I know what the issue was. LiteLLM expects ZAI_API_KEY and not ZHIPUAI_API_KEY which is what litellm_provider.py tries to se. nanobot/nanobot/providers/litellm_provider.py <https://github.com/HKUDS/nanobot/blob/d9d744d536aab152a57827a2931b894badde8479/nanobot/providers/litellm_provider.py#L52-L53> Lines 52 to 53 in d9d744d <http:///HKUDS/nanobot/commit/d9d744d536aab152a57827a2931b894badde8479> elif "zhipu" in default_model or "glm" in default_model or "zai" in default_model: os.environ.setdefault("ZHIPUAI_API_KEY", api_key) Yes，i solve this problem with ZAI_API_KEY.Here is my config. { "agents": { "defaults": { "workspace": "~/.nanobot/workspace", "model": "zai/glm-4.7", "maxTokens": 8192, "temperature": 0.7, "maxToolIterations": 20 } }, "providers": { "zhipu": { "apiKey": "", "apiBase": "https://open.bigmodel.cn/api/coding/paas/v4" } } } — Reply to this email directly, view it on GitHub <#2 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAYZWEKGR2EN2T7U3APXS34KC34PAVCNFSM6AAAAACTTNUVHWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNBRHE2TOMRUGA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

Anomit Ghosh

[Myskip]

Thanks but why would you post your API key publicly sir?!
On Tue, 3 Feb, 2026, 20:48 Myskip, @.***> wrote:
Myskip left a comment (HKUDS/nanobot#2)
<#2 (comment)>

@anomit https://github.com/anomit ah! I think I know what the issue was.

LiteLLM expects ZAI_API_KEY and not ZHIPUAI_API_KEY which is what
litellm_provider.py tries to se.

nanobot/nanobot/providers/litellm_provider.py

nanobot/nanobot/providers/litellm_provider.py

Lines 52 to 53 in d9d744d

	elif "zhipu" in default_model or "glm" in default_model or "zai" in default_model:
	os.environ.setdefault("ZHIPUAI_API_KEY", api_key)

Lines 52 to 53 in d9d744d
http:///HKUDS/nanobot/commit/d9d744d536aab152a57827a2931b894badde8479

elif "zhipu" in default_model or "glm" in default_model or "zai" in
default_model:
os.environ.setdefault("ZHIPUAI_API_KEY", api_key)

Yes，i solve this problem with ZAI_API_KEY.Here is my config.

{
"agents": {
"defaults": {
"workspace": "~/.nanobot/workspace",
"model": "zai/glm-4.7",
"maxTokens": 8192,
"temperature": 0.7,
"maxToolIterations": 20
}
},
"providers": {
"zhipu": {
"apiKey": "2e87684395d44155a48817fed9c8a877.N5Mq3tfAVojBtaG7",
"apiBase": "https://open.bigmodel.cn/api/coding/paas/v4"
}
}
}

—
Reply to this email directly, view it on GitHub
<#2 (comment)>, or
unsubscribe
https://github.com/notifications/unsubscribe-auth/AAAYZWEKGR2EN2T7U3APXS34KC34PAVCNFSM6AAAAACTTNUVHWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNBRHE2TOMRUGA
.
You are receiving this because you were mentioned.Message ID:
@.***>
Anomit Ghosh

fixed it, thanks

[geekapp2019]

我使用以下配置解决了：
"agents": {
"defaults": {
"workspace": "~/.nanobot/workspace",
"model": "zai/glm-4.7",
"maxTokens": 8192,
"temperature": 0.7,
"maxToolIterations": 20
}
},
"providers": {
"openai": {
"apiKey": "your key",
"apiBase": "https://open.bigmodel.cn/api/paas/v4/"
}
},

[pjperez]

Guys, the API key is still visible in the quoted text on #2 (comment) and #2 (comment)

Anyway, I've sent a PR with the second fix:
#95

[demoAlitalia]

i had the same problem on v0.1.3.post4。zai-glm4.7 use openai Protocol，
| nanobot.agent.loop:_process_message:138 - Processing message from cli:user